text-to-speech

This expert skill provides a robust framework for implementing high-quality, real-time text-to-speech (TTS) systems using the Kokoro TTS engine. Designed for developers building voice-enabled AI assistants like JARVIS, it emphasizes offline capabilities, low-latency streaming, and secure content handling. The skill guides you through the entire development lifecycle, from model configuration and voice selection to production-grade audio output and resource management, ensuring efficient GPU/CPU utilization without cloud dependencies.

• Kokoro TTS deployment and voice configuration for natural prosody and multi-voice support.

• Real-time streaming synthesis architecture to minimize latency in conversational interfaces.

• Security-focused audio generation, including input text filtering to block sensitive information and malicious payloads.

• TDD-first implementation workflows to verify synthesis quality, sample rates, and system stability.

• Performance optimization techniques such as audio chunking, model caching, and asynchronous execution to ensure smooth performance.

• Intended for developers working on local-first voice assistants, offline multimedia tools, or accessibility features requiring high-fidelity speech.

• Requires familiarity with Python, NumPy, SoundFile, and SoundDevice for hardware-level audio processing.

• Input requirements include plain text or SSML-formatted strings; output generates WAV format audio compatible with standard streaming buffers.

• Constraints include strict input validation to prevent DoS attacks via excessive text length and secure file system cleanup to manage temporary audio buffers effectively.

• Follow security practices to ensure that personal identifiable information (PII) is not accidentally synthesized or stored in logs during testing or production cycles.